Roadside acoustic sensors to support vulnerable pedestrians via their smartphone

TL;DR

This paper introduces a roadside acoustic sensor system that classifies vehicle types to warn vulnerable pedestrians via their smartphones, enhancing safety through real-time risk assessment.

Contribution

It presents a novel roadside acoustic sensor system using MFCC and LPC features with high-accuracy neural network classification for pedestrian safety.

Findings

MLP neural network achieves at least 96.77% accuracy

System effectively classifies vehicle types for risk assessment

Warning alarms are successfully sent to pedestrians' smartphones

Abstract

We propose a new warning system based on smartphones that evaluates the risk of motor vehicle for vulnerable pedestrian (VP). The acoustic sensors are embedded in roadside to receive vehicles sounds and they are classified into heavy vehicle, light vehicle with low speed, light vehicle with high speed, and no vehicle classes. For this aim, we extract new features by Mel-frequency Cepstrum Coefficients (MFCC) and Linear Predictive Coefficients (LPC) algorithms. We use different classification algorithms and show that MLP neural network achieves at least 96.77% in accuracy criterion. To install this system, directional microphones are embedded on roadside and the risk is classified there. Then, for every microphone, a danger area is defined and the warning alarms have been sent to every VPs smartphones covered in this danger area.